Did that study distinguish between drivers and non-drivers? I suspect that, if you exclude drivers, the risk of death does not increase with three or four or five drinks a day. Michael Savage last night said the British Medical Association defines moderate drinking as up to five drinks a day.
The study included every sort of death, including that of death while driving drunk. The ameliorative effects of two glasses a day on the heart overcome the relatively low incidence of drunk driving deaths at that level.
Interestingly, death increases at four drinks per day, compared to the tea-totaller. So, 1-3 per day leads to longer life, and two is optimal.
Ceteris paribus; If you're liver damaged by hepatitis already, not drinking is probably a good idea.